Direct Air Capture (DAC) Adsorbent Wheel Desiccant Cycle: Why Silica Gel Has Replaced Silica Gel and What Comes Next

What a DAC Adsorbent Wheel Actually Does

An adsorbent wheel in a Direct Air Capture plant is a rotating honeycomb drum that turns through two or three temperature-controlled zones. The honeycomb is typically 1.5-3.6 m in diameter with a face width of 0.2-1.0 m, and the channels are coated with a 200-300 micrometer thick layer of an amine-functionalised sorbent, a hygroscopic salt-impregnated silica gel, or a zeolite such as 13X HP. As the wheel turns at 5-20 revolutions per hour, every cell on the drum spends most of its time (60-80% of rotation) in the adsorption sector where ambient air at 415 ppm CO2 and 30-90% relative humidity is blown through the channels by a low-pressure fan (typical 200-500 Pa pressure rise).

The remaining 20-40% of the rotation is the desorption (regeneration) sector. Here a small, closed-loop stream of air or nitrogen is heated to 80-110 degrees C by an electrical heater, a heat-pump condenser, or low-pressure steam from a CHP unit, and blown counter-flow through the channels. This hot purge gas strips the bound CO2 (and co-adsorbed water in the hygroscopic case) and delivers a concentrated gas stream of 1-12 vol% CO2, depending on the sorbent material. That concentrated stream is the valuable product: it is dried, compressed, and either sequestered underground, mineralized into concrete aggregates, or upgraded for utilization in synthetic fuels or greenhouse horticulture.

What makes the wheel layout different from a fixed-bed TSA contactor is that the wheel provides continuous adsorption: there is no 2-6 minute cooling downtime between cycles because the desorption heat is applied to a different sector of the wheel at any given instant. The trade-off is parasitic fan power: a 2.5 m diameter wheel turning at 12 rpm with 70% of the face area in adsorption has roughly 4-7 m^2 of active adsorption area per drum, and the regeneration blower must move a fraction (8-20%) of the adsorption airflow at high temperature. Field measurements on Climeworks Mammoth class units put the regeneration-blower parasitic at 60-110 kWh per ton of CO2 captured, which is 2-3x the parasitic of a fixed-bed TSA blower. The full system-level energy trade-off is detailed in section 6 below.

The Silica Gel to Silica-Gel/LiCl Shift: Why It Happened 2020-2025

Straight silica gel was the default DAC sorbent between 2015 and 2019, and it remains the standard material for HVAC desiccant wheels used in dehumidification (40-80% RH, no CO2 target). But when applied to the DAC problem of capturing 415 ppm CO2 from atmospheric air, three failure modes became apparent. First, the equilibrium CO2 capacity of bare silica gel at 25 degrees C and 100 kPa is only 0.5-1.2 mmol/g, which translates to a wheel mass-loading of 2-5 kg per t/year installed capacity and a 8-15 ton wheel mass for a 1,000 t/year DAC plant - that is mechanically awkward and limits the maximum wheel diameter before bearings and shafts become a bottleneck. Second, silica gel has no chemical selectivity for CO2 over N2 or O2; it relies entirely on physisorption, and the CO2/N2 selectivity never rises above 8-12, so the regeneration loop can only deliver 0.5-1.5 vol% CO2 - the compressor downstream must work much harder to reach sequestration pressure. Third, desorption kinetics on silica gel demand 130-160 degrees C regeneration, which raises thermal duty and rules out direct use of low-grade waste heat (60-90 degrees C) from industrial processes.

The breakthrough that made third-generation wheels economically viable was the discovery that impregnating the silica-gel pore structure with a hygroscopic salt - lithium chloride in the leading commercial implementation, but calcium chloride and potassium formate are documented alternatives - drops the regeneration temperature to 70-95 degrees C while raising the apparent CO2 working capacity to 2.5-3.8 mmol/g at 400 ppm CO2. The mechanism is dual: the salt acts as a dispersed hygroscopic medium that pre-concentrates water on the surface (creating local micro-environments of 60-90% RH even when the bulk airflow is at 30% RH), and the resulting salt-water interaction forms a transient carbonate-like binding site that chemically adsorbs CO2 with an enthalpy of -50 to -75 kJ/mol, intermediate between physisorption (-25 kJ/mol) and full amine chemisorption (-80 to -100 kJ/mol).

The economic outcome is decisive. Field data published by Climeworks in 2024 and Heirloom in 2023 on their respective Mammoth and second-generation wheels show 35-55% lower thermal-energy duty than bare silica gel, 60-70% lower than straight amine-functionalised silica, and parity with first-generation liquid-amine MEA contactors at the capex line. The LiCl loading is 12-22 wt% of the silica-gel dry mass, well above the deliquescence point of bulk LiCl (which would be 47 wt%) so the salt remains dispersed in the mesoporous network rather than forming bulk brine droplets. Total material cost for a LiCl-impregnated silica-gel coating is roughly USD 9-14 per kg of finished coating at 100 MT/year scale, of which LiCl is USD 5-9, silica gel substrate USD 3-5; this is one-tenth of the cost of MOF-303 per kg at current commercial-scale pricing.

Inside the Desiccant Cycle: 3 Steps and 12 rpm

As the adsorbent wheel rotates, each channel on the drum experiences three discrete phases - adsorption, regeneration, and cooldown - in turn. The geometry is set by sector-plate baffles in the wheel housing that separate the adiabatic adsorption zone from the heated desorption zone; a typical split is 75% adsorb, 20% desorb, 5% cooldown, though high-RH regions sometimes use 80/15/5 to give the regeneration sector more residence time. The rotation speed, the sector splits, and the regeneration temperature are the three degrees of freedom a plant operator can adjust to tune the wheel for local climate and electricity price.

The cycle has a sharp optimum speed rather than a monotonic relationship. At less than 4 rpm the leading edge saturates with CO2 while the trailing edge still has unused capacity, producing 25-40% channeling loss. Above 25 rpm the desorption sector cannot fully strip the surface because the heat front has not had time to penetrate the 200-300 micrometer coating; regeneration efficiency drops from 95 to 65%, and partially-loaded sorbent circles back to the adsorption sector. The optimum for a 2.5 m diameter wheel at ambient adsorption and 80-110 degrees C desorption is 10-15 rpm, giving 4-6 seconds of contact in each sector with a 6-9% unused-capacity margin. Variable-frequency drives on the wheel motor allow the speed to be nudged ±2 rpm based on downstream CO2 purity telemetry, yielding 8-12% energy savings across a diurnal ambient-temperature cycle.

For readers new to the desiccant-cycle vocabulary, a useful contrast is Molecular Sieve 13X for CO2 Capture: PSA vs Chemical Absorption TCO, which walks through the same adsorption-pressure-swing trade-off in a fixed-bed layout, and Why Your Regeneration Cycle Keeps Getting Longer, which documents the same heat-front slow-down as a failure mode when the bed is too deep for the available regeneration temperature.

Relative Humidity as the Master Variable

For a LiCl-silica gel wheel, ambient RH is not just one variable among many - it is the variable that determines whether the wheel works at all. Below 30% RH (arid desert climate: Atacama, southwestern US, the Sahara fringe), the LiCl-silica gel's CO2 working capacity drops to 1.5-1.9 mmol/g because the salt-water synergy needs water pre-adsorbed on the silica-gel surface. A 13X-coated wheel in the same climate outperforms: 13X achieves 2.0-2.4 mmol CO2/g at 20% RH because the strong electrostatic field at the extra-framework cation (X^+ or Ca^2+) binds CO2 directly through quadrupole-cation interaction, and the comparatively weak competitive water adsorption at low RH does not displace CO2 from the cation site.

Between 40-70% RH (temperate oceanic climate: most of Europe, the Pacific Northwest, the eastern seaboard of North America, Japan, Korea), LiCl-silica gel is at its best. The surface water activity lands in the 0.4-0.7 range where the LiCl-water complex is most efficient at mediating CO2 binding, and the CO2 working capacity rises to 2.8-3.4 mmol/g. The 13X wheel in the same RH range saturates with co-adsorbed water and loses 30-40% of its CO2 capacity to that competitive adsorption; the regeneration temperature must be raised from 140 to 180 degrees C, which doubles thermal duty and negates the 13X capacity advantage.

Above 80% RH (tropical or monsoon climate), neither material works well in a wheel layout. The wheel channel becomes a wet surface where water vapor condenses in the mesopores and forms a thin brine film that blocks CO2 diffusion. A pre-dehumidification stage is mandatory: typically a 3A molecular sieve or a secondary desiccant-cooling wheel brings the air down to 50-60% RH before it enters the DAC wheel. This pre-dehumidification unit adds 400-600 kWh thermal per t-CO2 captured, and that overhead is one of the key reasons third-generation DAC projects in hot-humid regions such as Singapore have not adopted wheel contactors and instead use liquid-amine solvent scrubbing.

Sorbent Material Trade-Off: LiCl-Silica Gel vs 13X HP vs MOF-303 vs AER

For those designing a DAC wheel, the choice of adsorbent chemistry is dominated by ambient RH and on-site heat-integration temperature. The Molecular Sieve 4A vs 5A for PEM Electrolyzer Stack Recirculation post walks through the same 4A-vs-5A-vs-13X trade-off in a hydrogen-drying context and is a useful neighbour to read alongside this section.

13X HP grade as the arid-climate specialist

13X in the High-Purity (HP) grade, also marketed as LiLSX (lithium-exchanged low-silica-X), delivers the highest CO2 working capacity per gram among commercial zeolites at ambient temperature. The binder content must be below 5 wt% to keep pore mouths open; standard 13X binder-phenolic formulations have 15-20 wt% binders and lose 40-60% of the CO2 capacity when crushed into a 200-micrometer coating. The trade-off is that 13X HP has the highest regeneration temperature (140-180 degrees C) and a more energy-intensive heat-up cycle, but the higher CO2 purity in the regeneration loop (3-12 vol%) means a smaller downstream compressor and lower parasitic power. 13X wheels are therefore favoured in arid (<30% RH) regions and in plants with on-site waste heat (geothermal, biogas CHP, cement kiln shell-side cooling).

MOF-303 and the upper bound on working capacity

MOF-303 (Al(OH)(H2O)(1,3,5-benzenetricarboxylate), first synthesised by Yaghi and co-workers at UC Berkeley in 2018 and now in commercial pilot at BASF, NuMat, and Profmof) is the first metal-organic framework with an open metal site that binds CO2 at 60-85 degrees C regeneration temperature - lower than the regeneration temperature of LiCl-silica gel. The CO2/N2 selectivity at 1000 mbar partial pressure is 165, the highest of any commercial DAC sorbent, giving an 8-15 vol% regeneration-loop CO2 purity that halves downstream compression duty compared to LiCl-silica gel. The catch is cost: USD 80-150/kg today, projected to fall to USD 20-40/kg at 100 MT/year commercial scale by 2027-2028 once Atomics, NuMat, and Profmof commission their tonne-scale synthesis reactors.

Anion-exchange resin as the budget alternative

Quaternary-ammonium functionalised macroreticular anion-exchange resins (Diaion WA20, Lewatit MP500, Dowex Marathon 4000) are the lowest-cost DAC-grade sorbent at industrial CO2 concentrations (3-15 vol% in flue gas), but the CO2 working capacity at the DAC-relevant 415 ppm falls to 0.6-1.1 mmol/g, below the wheel-economic threshold of 1.5 mmol/g where the wheel mass becomes acceptably small for a 1,000-5,000 t/year plant. AER remains the workhorse for biogas upgrading (CO2 at 30-45 vol%) and indoor-air quality applications, where the CO2 partial pressure is 100-1,000x higher than ambient air.

Specific Energy Duty and the Parasitic-Load Story

Thermal energy duty for a LiCl-silica gel wheel contactor at design conditions (50% RH, 25 degrees C ambient, 90 degrees C regeneration) lands between 1,500 and 2,200 kWh thermal per metric ton of CO2 captured, or 1.5-2.2 GJ/t. This is comparable to Monoethanolamine (MEA) liquid scrubbing at the 3.0-4.0 GJ/t figure but achieved at ambient pressure and without the solvent-degradation penalties that liquid amines suffer. Electrical energy for fans and the regeneration heater is 350-550 kWh/t CO2, dominated by the regeneration-loop blower (typical 2,000-4,000 Pa pressure rise across the 70% open-face honeycomb at 5-15% of the adsorption flow rate).

The combined primary energy at a renewable-only electricity mix sits between 900 and 1,500 kWh per ton CO2 for the LiCl-silica gel wheel. At a coal-fired electricity mix, the equivalent primary energy rises to 1,900-2,900 kWh/t because of the 35-45% conversion efficiency of thermal power plants. The MOF-303 wheel projection falls 25-40% below the LiCl-silica gel figure; a 1 Mt/year DAC plant running on a 50% renewable grid would consume on the order of 1.5 TWh/year, equivalent to the output of a 200 MW dedicated wind or solar farm.

Working Life: Attrition, Salt Loss, and SO2 Poisoning

Service life is 5-9 years for LiCl-silica gel composite wheels and 4-7 years for 13X-coated wheels in continuous DAC operation. The dominant end-of-life mechanism is the attrition of the adsorbent coating from the rotating honeycomb. Each revolution exposes the coating to 12-40 seconds of channel-through airflow at 1-3 m/s face velocity, and the silica-gel or zeolite particles are progressively rounded and ejected as fines that must be filtered out of the regeneration loop. For comparison, Acid Gas Breakthrough in Molecular Sieve 13X walks through the same attrition mechanism in a fixed-bed 13X polishing train and the remedies available for SO2/H2S poisoning. Industry has settled on a 0.3-0.5% mass-loss-per-year rate for LiCl-silica gel; the coating falls below the 80% mass threshold after 6-8 years and the wheel must be recoated (sand-blast the old coating, re-impregnate, and oven-cure at 180 degrees C for 12 hours) or replaced.

13X-coated wheels are more attrition-resistant (0.15-0.25% mass loss per year) but accumulate trace SO2 and NO2 surface poisoning at 200-300 ppm/year in industrial airshed deployments, dropping the CO2 working capacity by 1-2% per year. The poisoning is reversible in early stages: dilute acid wash (0.5% HCl, 60 degrees C, 2-hour soak) followed by deionized-water rinse and thermal regeneration at 220 degrees C recovers 90-95% of initial capacity. For a remote-arid DAC plant far from industrial point sources, the SO2 poisoning is essentially zero and the 13X wheel achieves the upper 7-year service life. Coastal and near-power-plant deployments will see the lower 4-5 year service life and require periodic acid wash regeneration.

MOF-303 wheels are still under accelerated-aging study; preliminary data at 2,000-hour test cycles at 85 degrees C regeneration temperature and 90% RH indicates 4-5 year projected life, consistent with MOF amorphous-to-crystalline transition kinetics. The transition is not failure per se - the crystalline form has 60-70% of the fresh-MOF CO2 working capacity - but is a slow degradation that the operator tracks by monthly CO2 capture-rate benchmarking. Resin-based wheels (AER) are the shortest-lived at 2-4 years, limited by oxidative degradation of the quaternary nitrogen at the 110-140 degrees C regeneration temperature required to fully release CO2 from the strong base sites.

Acceptance Test Recipes for a New DAC Wheel

Three complementary field tests are the standard for accepting a new DAC wheel installation. The CO2 breakthrough sweep test places a CO2 sensor (NDIR, 0-2,000 ppm range, ±5 ppm accuracy) at the wheel outlet and samples at 1 Hz across a 24-hour period, integrating the outlet minus inlet CO2 concentration times the airflow to compute the average capture rate in kg-CO2/day per m^2 of wheel face area. A healthy LiCl-silica gel wheel delivers 8-15 kg-CO2/day/m^2 at sea-level airflow and 25 degrees C ambient; a healthy 13X wheel delivers 12-22 kg-CO2/day/m^2 in arid conditions. Values within ±10% of nameplate acceptance baseline pass; values below 75% trigger commissioning diagnostic action. The procedure shares many instruments with the field tests described in Ammonia Synthesis Loop Purge Gas Molecular Sieve Bed and Carbon Capture Solvent Regeneration Column Molecular Sieve Polishing, where a similar FTIR / NDIR / DP stack is standard at the polishing-bed outlet.

Temperature-swing verification is the second test. The regeneration-loop inlet temperature is stepped from 70 to 110 degrees C in 5-degree increments (for LiCl-silica gel) or 130 to 180 degrees C (for 13X HP), and the CO2 working capacity is plotted against temperature. Deviation from the expected linear curve indicates coating attrition, salt migration away from the adsorption zone, or thermal damage to the binder. The third test is pressure-drop baseline: the wheel is run at design airflow with both sectors open and the pressure drop across the wheel is compared to the manufacturer's nameplate value (typically 80-150 Pa for a 0.5 m deep wheel). A 30% increase over 12 months signals honeycomb channel fouling or coating densification and triggers cleaning or replacement action.

For each of these three tests, the field acceptance data should be logged weekly in the plant historian, with an alarm triggered at 75% of the original acceptance baseline for the CO2 capture rate, 110% of the original baseline for the pressure drop, and any deviation outside ±5 degrees C of the expected regeneration temperature curve. Acceptance tests should be repeated at 3, 6, 12, and 24 months post-commissioning, then annually.

Procurement Specification Template (10 Lines You Need)

When issuing a vendor RFQ for a DAC wheel, the seven acceptance criteria above are the line items most often negotiated. LiCl content specification is the single most common dispute: vendors sometimes quote nominal LiCl loading at 22 wt% on as-synthesized substrate but deliver coated wheels at 14-16 wt% after the calcination step. Always require the LiCl loading test (line 7) as a separate pass/fail line item rather than tying it to a vendor's "nominal" claim.

Next Steps

If you are sizing a new DAC wheel contactor or evaluating a sorbent swap (LiCl-silica gel to MOF-303, or 13X wheel to a 4-bed TSA contactor), the fastest path is a 30-minute engineering call with our team. Send us the seven specification parameters from the table above plus your ambient RH range (10th, 50th, 90th percentile), your on-site heat-integration temperature (geothermal, CHP jacket-water, kiln shell side), and your annual capture target (t/year) - and we will return a sorbent selection recommendation, a wheel diameter and rotation-speed recommendation, and a 5-year operating-cost projection within 48 hours.

Frequently Asked Questions